Category-Specific Deficits and Implications for the Structure of Conceptual Knowledge David Price

Basic Facts/Trends Distinguishing features of category-specific deficits: Distinguishing features of category-specific deficits: 1. Patients seemingly exhibit disproportionate and selective impairment of knowledge of one category of objects v. other categories 2. Most studies of instances of category-specific deficits involve sparing or selective impairment of ‘living things’, ‘non-living things’, ‘animals’, ‘fruits/vegetables’, and ‘artifacts’ 3. Unclear as to whether instances of category-specific deficits are a result of damage to a categorically-organized semantic system, or damage to specific modality of knowledge (e.g. visual or functional)

The Basic Motivation Scientific explanation of category-specific deficits center around the basic question of Q1: How is conceptual knowledge organized? Scientific explanation of category-specific deficits center around the basic question of Q1: How is conceptual knowledge organized? Cognitive Neuropsychology attempts to answer Q1 by examining cases of individuals who seem disproportionately impaired at naming, defining, and/or characterizing members of a given category, and do not exhibit similar deficits in other categories. (e.g. S can’t name any x’s that are animals, but can name all other x’s of category ) Cognitive Neuropsychology attempts to answer Q1 by examining cases of individuals who seem disproportionately impaired at naming, defining, and/or characterizing members of a given category, and do not exhibit similar deficits in other categories. (e.g. S can’t name any x’s that are animals, but can name all other x’s of category ) If cognitive scientists can accurately predict the cause and nature of cases of category-specific impairments, then a theoretical backdrop for a cognitive theory of conceptual knowledge could emerge If cognitive scientists can accurately predict the cause and nature of cases of category-specific impairments, then a theoretical backdrop for a cognitive theory of conceptual knowledge could emerge

Methodology of this presentation 1. Give brief summary and history of category- specific deficits 2. Discuss two competing theories of category- specific deficits (Sensory/Functional Theory v. Domain-Specific Theory) 3. Discuss these in terms of particular cases which purported to support them 4. Discuss explanatory inconsistencies for both of the theories, discuss soundness of their respective presumptions

History of the Condition First clear example: Warrington & Shalice (1984) describe 4 patients recovering from herpes simplex encephalitis that appear disproportionately impaired at both comprehension and naming of living things v. nonliving things First clear example: Warrington & Shalice (1984) describe 4 patients recovering from herpes simplex encephalitis that appear disproportionately impaired at both comprehension and naming of living things v. nonliving things W & S (1984) also observed case of brain-damaged individual w/reverse dissociation (i.e. nonliving things were problematic, but living things were not) W & S (1984) also observed case of brain-damaged individual w/reverse dissociation (i.e. nonliving things were problematic, but living things were not) W & S inferred from these 5 instances that selective brain damage could result in category-specific semantic deficits W & S inferred from these 5 instances that selective brain damage could result in category-specific semantic deficits Thus, a study of the nature of category-specific semantic deficits would prove fruitful for the study of the structure of semantic knowledge in normal subjects Thus, a study of the nature of category-specific semantic deficits would prove fruitful for the study of the structure of semantic knowledge in normal subjects

History of the Condition (con’t) All 4 cases showed disproportionate difficulty for living vs. nonliving things All 4 cases showed disproportionate difficulty for living vs. nonliving things J.B.R. reported as unable to recognize/name 2/48 living things (animals or plants) but could describe/named 45/48 nonliving things J.B.R. reported as unable to recognize/name 2/48 living things (animals or plants) but could describe/named 45/48 nonliving things S.B.Y. identified 0/48 living things but identified 36/48 nonliving things S.B.Y. identified 0/48 living things but identified 36/48 nonliving things Subsequently, confirmation of dissociation b/w living v. nonliving things obtained in many other studies of brain-damaged subjects (variance in etiology; herpes simplex encephalitis, trauma, cerebro- vascular accident, neurosurgery) Subsequently, confirmation of dissociation b/w living v. nonliving things obtained in many other studies of brain-damaged subjects (variance in etiology; herpes simplex encephalitis, trauma, cerebro- vascular accident, neurosurgery) Reverse dissociation also confirmed: impaired performance nonliving objects v. spared ability with living things Reverse dissociation also confirmed: impaired performance nonliving objects v. spared ability with living things

Basic Methodological Assumptions/Practices Picture-naming task necessarily involves activation of semantic information about a particular object (usual method is to use Snodgrass and Vanderwart picture set) Picture-naming task necessarily involves activation of semantic information about a particular object (usual method is to use Snodgrass and Vanderwart picture set) Comprehension tests: administered after picture-naming task, these tests involve the individual’s ability to distinguish features of objects; are meant to demonstrate either: Comprehension tests: administered after picture-naming task, these tests involve the individual’s ability to distinguish features of objects; are meant to demonstrate either: (a) that the individual’s problem results from the inability to recognize the problem-object, in which case the relevant system is not at all presented with the necessary features to produce the desired output -OR- (b) That the individual’s problem results from an inability to properly extract properties of the problem-object which are somewhere present in the system (the latter seems more likely, given the lack of instances in which a subject’s dissociation is necessary; i.e. it is not the case that S cannot properly identify all entities of category c, it is that he cannot properly identify all entities in category c effectively, or consistently, or functionally)

Is the semantic system organized categorically? Hypothesis that the semantic system is categorically organized initially not seriously entertained Hypothesis that the semantic system is categorically organized initially not seriously entertained Inconsistent findings for category hypothesis from initial cases usually manifest in that patterns of affected semantic categories failed to correspond to definite categorical distinctions Inconsistent findings for category hypothesis from initial cases usually manifest in that patterns of affected semantic categories failed to correspond to definite categorical distinctions Reasons for initial rejection: Reasons for initial rejection: 1. If semantic system is categorically organized, then JBR’s category- specific deficit should correspond to definite semantic categorical distinctions, like living v. nonliving 2. JBR’s deficit is in animal & plant categories (living) AND food category (nonliving) 3. Thus, hypothesis that semantic system is categorically organized is violated

Proposal of Sensory/Functional Theory of Category- specific deficits Refutation of hypothesis that semantic system is categorically organized lead Warrington & Shalice (1984) to propose Sensory/Functional theory Refutation of hypothesis that semantic system is categorically organized lead Warrington & Shalice (1984) to propose Sensory/Functional theory This theory says that JBR and SBY’s case can be explained by accounting for damage to visual semantic subsystem and/or damage to functional semantic subsystem This theory says that JBR and SBY’s case can be explained by accounting for damage to visual semantic subsystem and/or damage to functional semantic subsystem Visual semantic subsystem: damage to this area results in disproportionate deficit of living things and foods because the identification of both contingent on visual features Visual semantic subsystem: damage to this area results in disproportionate deficit of living things and foods because the identification of both contingent on visual features Functional semantic subsystem: damage to this area results in disproportionate deficit of nonliving things because nonliving things are distinguished based on utility or function. Functional semantic subsystem: damage to this area results in disproportionate deficit of nonliving things because nonliving things are distinguished based on utility or function. THUS, apparent categorical nature of semantic deficits doesn’t reflect categorical structure of semantic knowledge, but reflects instead a more basic organizing principle of semantic system contingent on sensory and functional cues THUS, apparent categorical nature of semantic deficits doesn’t reflect categorical structure of semantic knowledge, but reflects instead a more basic organizing principle of semantic system contingent on sensory and functional cues

Sensory/Functional Theory Principle Principle 1. Non-categorical, modality-specific organization 2. Differential attribution of living v. nonliving things Explanation 1. Semantic system organized into modality-specific subsystems (visual/perceptual, functional/associative) 2. Ability to recognize/name living things contingent on visual/perceptual info.; non- living things contingent on functional/associative info.

Case Study 1: Bert (Barry & McHattie) Barry & McHattie case study supports Sensory/Functional Theory Barry & McHattie case study supports Sensory/Functional Theory Bert, age 74, had stroke which caused right visual field defect, no problems with conversation, intellectually intact, preserved comprehension Bert, age 74, had stroke which caused right visual field defect, no problems with conversation, intellectually intact, preserved comprehension Complained of both word-finding difficulties and pronounced inability to name entities in the category of animals Complained of both word-finding difficulties and pronounced inability to name entities in the category of animals Barry and McHattie wanted to explore possibility of category-specific anomia for animals Barry and McHattie wanted to explore possibility of category-specific anomia for animals Initial testing: appeared to have intact object recognition (verified by appropriate use of household items and by his performance of a test which tested object recognition; shown 3 different pictures, had to indicate by pointing which of the three pictures had objects which contained items with the same name, otherwise 2 things of same name but different physical appearances) Initial testing: appeared to have intact object recognition (verified by appropriate use of household items and by his performance of a test which tested object recognition; shown 3 different pictures, had to indicate by pointing which of the three pictures had objects which contained items with the same name, otherwise 2 things of same name but different physical appearances)

Bert con’t Picture naming task: Bert presented with 100 line drawings of Snodgrass and Vanderwart test Picture naming task: Bert presented with 100 line drawings of Snodgrass and Vanderwart test Either produced correct name of the picture, or indicated he couldn’t say it and would subsequently partake in some degree of description Either produced correct name of the picture, or indicated he couldn’t say it and would subsequently partake in some degree of description For instances in which Bert could not name it, he indicated too that he ‘knew the thing’. For instances in which Bert could not name it, he indicated too that he ‘knew the thing’.

Results of Picture Naming Success rate: Success rate: (a) Animal naming=12% (b) Fruits/vegetables naming=47% (c) Musical instruments= 50% (d) Tools=60% (e) Vehicles=67% (f) Articles of clothing=71% (g) Body-parts=100% Worst at naming animals as compared to any other category Data does not fit hypothesis that semantic knowledge organized into discrete categories Word familiarity, visual complexity of pictures, and age-of-acquisition did effect Bert’s responses for the task (e.g. Bert was better at naming familiar items vs. non- familiar)

Possible Reasons for Bert’s Deficit Reasons (a) Inability to recognize animals (b) Problematic integrity of semantic representation of animal, or problem with the utilization of those representations which support proper naming Explanation (a) refuted, Bert could point to appropriate animal in a picture of two related animals when prompted by an animal name (b) Possible, comprehension test follows to seek out this possibility

Comprehension Test 4 questions for each of the 30 animals + 30 items presented to Bert: 4 questions for each of the 30 animals + 30 items presented to Bert: One required ‘yes’ response about category of the item (e.g. Is a squirrel an animal?) One required ‘yes’ response about category of the item (e.g. Is a squirrel an animal?) One contained false information about category of object (e.g. is a mushroom a musical instrument?) One contained false information about category of object (e.g. is a mushroom a musical instrument?) One was a correct question about physical appearance of item of category (e.g. has a squirrel got a bushy tail?) One was a correct question about physical appearance of item of category (e.g. has a squirrel got a bushy tail?) One was false question about physical appearance (e.g. has a squirrel got wings?) One was false question about physical appearance (e.g. has a squirrel got wings?)

Results of Comprehension Test Type of Question Type of Question 1) Category Questions 1) Category Questions 2) Physical Property Questions 2) Physical Property Questions Results 1) 93.3% correct for category questions 2) 76.7% correct for questions about physical properties **Thus, Bert’s inability to name and recognize members of the animal category is because of a malfunctioning visual semantic subsystem, according to Sensory/Functional Theory *significant difference b/w 1 and 2, thus his category-specific problem is a reflection of a failure to account for physical properties which distinguish members of animal category from members of other categories

Problems for the Sensory/Functional Theory Because this theory assumes that ability to recognize all living things differentially contingent on info internal to the same visual/perceptual subsystem, prediction is that a dissociation will not be observed within same category ‘living things’. Because this theory assumes that ability to recognize all living things differentially contingent on info internal to the same visual/perceptual subsystem, prediction is that a dissociation will not be observed within same category ‘living things’. CONTRARY to this is the report of individuals with disproportionate deficits for ‘fruits/vegetables’ v. ‘animals’ CONTRARY to this is the report of individuals with disproportionate deficits for ‘fruits/vegetables’ v. ‘animals’ Predicts patients will present w/disproportionate deficits for type of info upon which successful recognition of impaired category is assumed to depend on. Early reports supported this, but data has been criticized on methodological grounds (pictures unmatched for word frequency et cetera) Predicts patients will present w/disproportionate deficits for type of info upon which successful recognition of impaired category is assumed to depend on. Early reports supported this, but data has been criticized on methodological grounds (pictures unmatched for word frequency et cetera) Predicts association b/w disproportionate deficit for type of knowledge & disproportionate deficit for category of objects that depends on that knowledge. Predicts association b/w disproportionate deficit for type of knowledge & disproportionate deficit for category of objects that depends on that knowledge. CONTRARY to this: patients reported w/ disproportional deficit in functional/associative knowledge but no associated disproportionate deficit for living things v. non-living things CONTRARY to this: patients reported w/ disproportional deficit in functional/associative knowledge but no associated disproportionate deficit for living things v. non-living things

Domain-Specific Hypothesis Central assumption: evolutionary pressures resulted in specialized neural circuits dedicated to processing (both conceptually and perceptually) distinct categories of objects. Instantiates the notion that semantic system may be organized categorically. (Warrington, 1981) Central assumption: evolutionary pressures resulted in specialized neural circuits dedicated to processing (both conceptually and perceptually) distinct categories of objects. Instantiates the notion that semantic system may be organized categorically. (Warrington, 1981) Provides independent specifications for how we can characterize a conceptual category; restricted only to categories that were advantageous with respect to survival and reproductive advantages Provides independent specifications for how we can characterize a conceptual category; restricted only to categories that were advantageous with respect to survival and reproductive advantages

Domain-Specific Hypothesis con’t Predictions: Predictions: 1. If there are distinct neural systems dedicated to categories of ‘animals’, ‘fruits/vegetables’, and possibly ‘tools’, then it’s impossible for function of one system, if damaged, to be recovered by other systems (i.e. predicts poor-recovery of impaired performance 2. No necessary correspondence b/w deficit for knowledge of modality x and conceptual deficit for category of objects 3. Patients might present category-specific visual agnosia. This follows from assumption that perceptual stages of object recognition are possibly functionally organized according to domain-specific restrictions. (supported by observation of subjects with equal deficits to visual and functional knowledge of living things, but visual agnosia for living things v. non-living things)

Case 2: Patient EW (Caramazza and Shelton, 1998) This case is purported to support the Domain-Specific Hypothesis (Caramazza and Shelton, 1998) This case is purported to support the Domain-Specific Hypothesis (Caramazza and Shelton, 1998) Patient EW: selective deficit for category of animals, equally impaired for visual and functional attributes of members of animal category v. other living things and artifacts Patient EW: selective deficit for category of animals, equally impaired for visual and functional attributes of members of animal category v. other living things and artifacts E.W. tested via Picture Naming, Sound Identification, Object decision, Parts decision, Visual processing, and Central-Attribute judgments E.W. tested via Picture Naming, Sound Identification, Object decision, Parts decision, Visual processing, and Central-Attribute judgments

Picture Naming Snodgrass & Vanderwart picture set, matched for familiarity and frequency Snodgrass & Vanderwart picture set, matched for familiarity and frequency EW disproportionately impaired at naming animals (55%) v. non-animals (82%). Controls were 100% at naming animals and 98% for non-animals EW disproportionately impaired at naming animals (55%) v. non-animals (82%). Controls were 100% at naming animals and 98% for non-animals EW’s performance both quantitatively and qualitatively different for animals v. non-animals. Animals, EW either named picture incorrectly or did not recognize picture; for non-animals, EW recognized the picture but couldn’t retrieve name EW’s performance both quantitatively and qualitatively different for animals v. non-animals. Animals, EW either named picture incorrectly or did not recognize picture; for non-animals, EW recognized the picture but couldn’t retrieve name EW’s deficit did not extend to other living things like the category ‘fruit/vegetables’ (performed very well in this category) EW’s deficit did not extend to other living things like the category ‘fruit/vegetables’ (performed very well in this category)

Sound Identification Task presented E.W. with 32 characteristic sounds of animals and 32 non-animal sounds Task presented E.W. with 32 characteristic sounds of animals and 32 non-animal sounds E.W. impaired at naming animals v. non-animals based on characteristic sound (25% correct v. 63%) E.W. impaired at naming animals v. non-animals based on characteristic sound (25% correct v. 63%) This indicates that naming impairment is not restricted to one sensory system This indicates that naming impairment is not restricted to one sensory system

Object Decision Asked to decide ‘yes’ or ‘no’ whether depicted object was real. Asked to decide ‘yes’ or ‘no’ whether depicted object was real. Performance on this task is interpreted as reflecting integrity of visual/structural description system (this was system purported to be one of the causes of Bert’s naming problems) Performance on this task is interpreted as reflecting integrity of visual/structural description system (this was system purported to be one of the causes of Bert’s naming problems) Significantly below normal range for differentiating b/w real from unreal animals (60% correct, controls were 90%) Significantly below normal range for differentiating b/w real from unreal animals (60% correct, controls were 90%) Within normal range for distinguishing real from unreal non-animals (92% correct, controls were 84%) Within normal range for distinguishing real from unreal non-animals (92% correct, controls were 84%)

Parts Decision Task EW to decide which of two heads went with headless body EW to decide which of two heads went with headless body Severely impaired on this task for animals (60%, controls were 100%) Severely impaired on this task for animals (60%, controls were 100%) Normal range for artifacts (97%, controls were 97% Normal range for artifacts (97%, controls were 97% Indicates that EW doesn’t have deficit for visual processing for complex stimuli Indicates that EW doesn’t have deficit for visual processing for complex stimuli Suggests her impairment for objects reality decision for animals is categorically based Suggests her impairment for objects reality decision for animals is categorically based

Central-Attribute Judgments EW asked to decide if a given attribute was true of a given item. This tests ability to distinguish properties/features of an object EW asked to decide if a given attribute was true of a given item. This tests ability to distinguish properties/features of an object Severely impaired for attributes pertaining to animals (65%, controls were %) Severely impaired for attributes pertaining to animals (65%, controls were %) Normal range for attributes pertaining to non-animals (95%, controls were %) Normal range for attributes pertaining to non-animals (95%, controls were %) Equivalently impaired for visual/perceptual and functional/associative knowledge of living things (65% for both) Equivalently impaired for visual/perceptual and functional/associative knowledge of living things (65% for both) Normal range for both types of knowledge for non-animals Normal range for both types of knowledge for non-animals Thus, EW’s performance on central-attribute questions implies that her deficit is not restricted to production Thus, EW’s performance on central-attribute questions implies that her deficit is not restricted to production

Conclusions and Speculations of EW Case Results summarized by Caramazza & Shelton: Results summarized by Caramazza & Shelton: 1. EW has category-specific deficit restricted to category of animate objects (this persisted under strict control of nuisance factors like familiarity, visual complexity, frequency, or combination of them) 2. Category-specific deficit manifests in visually and auditorily recognizing animate objects 3. Category-specific deficit in language comprehension; manifested in poor performance w/ statements about animate objects v. normal limits w/statements about other living things and artificats

Domain Specific Organization of Conceptual Knowledge and Evolutionary Adaptation Fact that the categories animals, fruits/vegetables, and artifacts can be independently impaired suggests that it is the case that ONLY these three categories form basis for conceptual organization of conceptual knowledge Fact that the categories animals, fruits/vegetables, and artifacts can be independently impaired suggests that it is the case that ONLY these three categories form basis for conceptual organization of conceptual knowledge This assumption requires independent empirical justification; recourse to evolutionary adaptation could provide such justification This assumption requires independent empirical justification; recourse to evolutionary adaptation could provide such justification Recognition of animal category would imply ability to respond quickly to all types of animals, including predators. Physiological evidence: we can detect movement of living organisms faster than movement of non-living things, we require less information Recognition of animal category would imply ability to respond quickly to all types of animals, including predators. Physiological evidence: we can detect movement of living organisms faster than movement of non-living things, we require less information Evolutionary adaptations for recognition of animals and plant life provides skeletal neural structures which would organize perceptual, conceptual, and linguistic knowledge modern humans have of these categories Evolutionary adaptations for recognition of animals and plant life provides skeletal neural structures which would organize perceptual, conceptual, and linguistic knowledge modern humans have of these categories

Domain Specific Organization of Conceptual Knowledge and Evolutionary Adaptation con’t Implication: only true category-specific deficits are those which involve categories of animal, plant-life, and possibly artifacts Implication: only true category-specific deficits are those which involve categories of animal, plant-life, and possibly artifacts Implication: supposition that specialized neural mechanisms are more likely to be selectively damaged implies that frequency of category- specific deficits for living things should be higher than non- living things. Implication: supposition that specialized neural mechanisms are more likely to be selectively damaged implies that frequency of category- specific deficits for living things should be higher than non- living things. More specific categories like tools would not constitute a category-specific deficit, because the category of tools does not serve the function of organizing semantic knowledge into categories This 2 nd implication is supported in that most cases of category-specific deficits are manifested in animal or plant categories (aka, fruits/vegetables)

Conclusions Given that, as it stands now, there is a higher degree of supporting evidence for domain-specific theory compared w/sensory-functional theory, domain-specific theory explains category-specific deficits better Given that, as it stands now, there is a higher degree of supporting evidence for domain-specific theory compared w/sensory-functional theory, domain-specific theory explains category-specific deficits better Also, much of the evidence for sensory/functional theory was thrown out due to fact that uncontrolled stimulus factors like familiarity could account for apparent deficit Also, much of the evidence for sensory/functional theory was thrown out due to fact that uncontrolled stimulus factors like familiarity could account for apparent deficit When these factors were controlled, many of the early cases supporting sensory/functional theory were shown to have been partially a function of uncontrolled stimulus factors When these factors were controlled, many of the early cases supporting sensory/functional theory were shown to have been partially a function of uncontrolled stimulus factors

Works Cited Caramazza, A., & Mahon, B. (2003). The Organization of Conceptual Knowledge: the Evidence from Category-Specific Semantic Deficits. Trends in Cognitive Sciences, 7: 8, pp Caramazza, A., & Mahon, B. (2003). The Organization of Conceptual Knowledge: the Evidence from Category-Specific Semantic Deficits. Trends in Cognitive Sciences, 7: 8, pp Caramazza A., & Shelton, J. (1998). Domain-Specific Knowledge Systems in the Brain: The Animate-Inanimate Distinction. Journal of Cognitive Neuroscience, 10:1, pp Caramazza A., & Shelton, J. (1998). Domain-Specific Knowledge Systems in the Brain: The Animate-Inanimate Distinction. Journal of Cognitive Neuroscience, 10:1, pp Lombardi, L. & Sartori, G. (2004). Semantic Relevance and Semantic Disorders. Journal of Cognitive Neuroscience, 16:3, pp Lombardi, L. & Sartori, G. (2004). Semantic Relevance and Semantic Disorders. Journal of Cognitive Neuroscience, 16:3, pp Warrington, E.K., & Shalice, T. (1984). Category-Specific Semantic Impairments. Brain, 107, Warrington, E.K., & Shalice, T. (1984). Category-Specific Semantic Impairments. Brain, 107,